Who is afraid of ChatGPT? Professor Benjamin Grewe on AI and its limits.

25. May 2023 – Katharina Schlangenotto

Benjamin Grewe takes things quite calmly. Artificial intelligence is his profession and he sees it more as a good friend than an evil enemy.

Artificial intelligence, he says, is still artificially generated, in programs that play something that has been fed to them. It is capable of learning to a limited extent, but: “The AI has never interacted in the world, never experienced things, never ridden a bus, and can therefore really only understand things based on texts.”

ChatGPT uses the GPT3 language model as its basis. It can generate human-like texts and the abbreviation stands for Generative Pre-trained Transformer 3. The chatbot was developed by OpenAI and is available free of charge during the test phase.

What is intelligence from Professor Benjamin Grewe’s perspective?

Intelligence, according to Benjamin Grewe, an accomplished neuroscientist and researcher, is for him “the ability to generate behavior in all kinds of situations that brings us closer to our goal. And a machine like ChatGPT can’t do that because it can only generate text.”

So far, AI lacks self-awareness, or the ability to make sense of experience. Grewe, who studied at Stanford University, among other places, explains it this way, “When we’re driving, for example, we see a ball roll onto the road. We then immediately know that there are children playing there and we have to be careful. We’ve learned that and apply the brakes. A car controlled by artificial intelligence that has not experienced this scene in this way does not know what to do here. It can’t make sense of the scene.”

So artificial intelligence can only produce what it is taught? So far, yes, but Benjamin Grewe dreams of more.

The professor, who was born in Schleswig-Holstein and now lives and researches in Zurich, would like to see artificial intelligence act like a virtual assistant. “AI needs to be developed to the point where it understands 100 percent of the tasks it is given. That’s not where we are today. I envision an AI that understands the world like we do and also acts similarly to us humans.”

Nothing to worry about? Really?

Phew, that does sound a bit scary. Benjamin Grewe sees no reason to worry, because he doesn’t see AI as a replacement for humans, but as a tool that can support humans in many areas. “I think you would have to have some kind of integrated intelligence that grows up in the world with us, maybe even goes to school,” he explains, saying at the same time that the current state of AI is far from surpassing human intelligence.

However: that’s where it is heading to. It is a question of time.  

So what about ChatGPT, the language tool that rightly scares the copywrighter scene?

“The potential of such AI systems is immense.”

Benjamin Grewe believes that the GPT has so far only been able to perform the simplest tasks because it has so far been limited to statistical learning algorithms. These calculate which word is most likely to match a given text.

But that is the state of the art today, because the potential of such AI systems is immense.

In an article for “Finanz und Wirtschaft – fuw,” he and other colleagues wrote: “The next milestone will not be GPT-4, five or even higher, but a bot that teaches itself human reasoning through its own journeys of discovery in virtual worlds. Especially in Switzerland, researchers are working on systems that can learn fully autonomously, fathom the world, and thus become more reliable helpers.”

And he adds, “We would do well to actively shape this next technological step at the forefront of the world, because the stakes are high. Switzerland, as the innovation world champion and global hotspot of the AI scene, is predestined for this.”

Professor Benjamin Grewe is one of the most exciting researchers of our time. In his innovative lectures, he shares his knowledge and passion about AI and its further development, in which he is significantly involved with a team of scientists.

Fear? He doesn’t know what that is. He says it’s much more important to actively contribute to a positive, ethical future. A future that helps people and supports them in life. Benjamin Grewe takes his listeners a step further into the future. And isn’t that something?

Prof. Benjamin Grewe

Neuroscientist & Researcher Artificial Intelligence